THIS IS A TEST INSTANCE. ALL YOUR CHANGES WILL BE LOST!!!!
Meeting Notes
Introduce teams and individuals wanting to participate
Organizations with attendees:
- Instaclustr
- DataStax
- Bloomberg
- New Relic
- Apple
- Orange
- Ippon Technologies
Discuss a framework for participation
- Outline needs to be created in documentation. CEP to be created (Ben and Patrick will take on first draft)
- Time zone considerations seem to be good. Two meetings will have to work for max coverage.
- Suggestion is for on-going bi-weekly meetings
Outline some early goals and next steps
- Find the common ground between all the operators
- Think of terms of pluggable layers to keep implementations flexible.
- Gather feedback from users and potential contributors.
- New Relic - “Just starting and look at operator options to bring our databases in k8s”
Discussion: What makes an operator production ready?
- Cross region to kubernetes clusters. Match C* capabilities
- Has to be run by the ops team not the dev team
- Daily tasks have to be easy
- Backup and Restore
- Maintenance tasks have to be easy
- Working with stateful sets
- Adding to dtest and part of testing
- Everything needs to be configurable
- Network stack in k8s
- DNS in Cassandra has to get fixed - Possible solution(https://github.com/kubernetes/kubernetes/issues/28969)
- Observability by default
Outstanding Questions
- What about the “official” docker image? Currently not managed by the Apache Cassandra project. Seems like a mis-match.
- Connection with sidecar project?
- What about the release process? Would it be tied to Cassandra releases?
- Committer process: Who would be approving PRs. Operator code is all in Go which would require a different group of committers for this project.
- Taking people on more committees may be the right choice
Is a Kubernetes Operator a ASF project or should it live somewhere else?
Zoom Video Recordings
| |
|